The manufacture of semiconductors typically includes a number of ion implant steps whereby a workpiece, usually a silicon wafer, is presented to an impinging ion beam. The velocity of the ions is set such that they will bury into the workpiece and come to rest at a desired depth, forming an implanted region or layer. The ion beam in almost all cases is smaller than the silicon wafer and measures must be taken to implant the entire wafer surface. The ion dose must be uniformly distributed over the wafer surface and preferably every ion will impinge the wafer surface at the same angle.
In a typical ion implanter, a relatively small cross-section beam of dopant ions is scanned relative to a silicon wafer, which can be done in essentially one of three ways: scanning of the beam in two directions relative to a stationary wafer; scanning of the wafer in two directions relative to a stationary beam; or a hybrid technique where the beam is scanned in one direction while the wafer is mechanically scanned in a second, typically orthogonal, direction. In all cases, the ability to impinge the wafer at selectable non-zero angles of incidence with minimum angle spread is required.
Each technique has advantages and disadvantages. A widely used approach has been to mount a batch of wafers on a disc or at the end of spokes on a rotating wheel which causes a fixed direction ion beam to impinge upon each wafer in turn. The rotating disc or wheel is then scanned to and fro and causes the fixed direction ion beam to impinge upon every part of the surface of every wafer. This technique has proven successful with smaller wafer sizes but is less attractive with today's larger wafers.
For implantation into larger (300 mm) wafers, batch processing is currently not preferred because the individual work-in-process value of each wafer introduces a significant financial risk, should a problem arise during implantation which causes a batch to be scrapped. Another reason batch processing is not preferable is because production flow is simplified for non-batch processing, especially for wafer lot sizes which do not match the batch number for a particular wheel. Still another reason is that single wafer implanters avoid implant angle errors that are inherent in wheel type batch implanters built to-date. Two-dimensional electrostatic or magnetic scanning of an ion beam in orthogonal directions relative to a stationary wafer is a process that has been implemented in early-generation commercial implanters, but modern implanters require all the ions to be traveling in as closely as possible parallel paths, which is increasingly difficult as wafer sizes grow (especially for two-dimensional scanning). Present single-wafer scanning techniques tend to employ so-called hybrid scanning, where the ion beam is scanned or formed into a stationary ribbon by electrostatic or magnetic means in a first direction that is perpendicular to the beam line axis in the ion implanter, and the wafer is mechanically moved in a second, generally orthogonal, direction. In each case, the apparatus to either scan and restore the beam to a parallel condition or mechanically scan the wafer have problematically grown very massive, expensive, and complex as wafer sizes have increased to the present 300 mm wafer size, and this problem will increase as wafer sizes increase in the future.
There are significant advantages (in terms of cost, footprint-minimization of the length of the beam line, weight, and lower mechanical complexity) if a much simpler, more compact and lower cost scanning system is employed. These advantages can be used to produce cost effective, application specific, implanter tools, which can be optimized for one or just a few implant process steps. In this way, the ever increasing complexity and performance compromises of broad range implanters is avoided. For example, one implanter may be optimized for light ions at moderate energies such as is required for high-dose hydrogen implanting for layer separation and another implanter may be optimized for low energy high-dose boron such as is used in source/drain or source/drain extension implants. Other examples could optimize performance for higher energy low-dose or cluster ion beams such as decaborane or GCIB (Gas Cluster Ion Beam).
U.S. Pat. No. 4,295,048 discloses a system of deflecting electrons to a plurality of discrete positions using a series of sequentially operated magnets arranged along a beam pipe. The electron beam is deflected by each magnet in turn to an electron window associated with each magnet. U.S. Pat. No. 6,617,586 also discloses a system for deflecting an electron beam, in this case a pulsed electron beam, to a plurality of discrete positions using a series of sequentially operated magnets. However, both of these systems are only suitable for deflecting electrons to discrete positions, and do not allow for continuous scanning of a charged particle beam to produce a uniform dosage across a target wafer. Furthermore, both systems are only described in relation to electron beams, and in the case of U.S. Pat. No. 6,617,586, specifically to pulsed electron beams.
It is therefore an object of the present invention to provide scanning of the beam of charged particles across a target surface in a continuous manner, to provide for uniform dosage of particles across the surface, or at least to provide the public with a useful choice.